On Tue, Dec 1, 2009 at 8:56 AM, Nathann Cohen <nathann.co...@gmail.com> wrote: > Hmmm... If we can make Sage's C graphs as fast as NetworkX's , I > assure you it will be very hard for them to compete with LP on our > side and so many features around in Sage... As most of NetworkX's > functions are not very hard to rewrite once the basis is set ( graph > structure, neighbors, etc ) I am more set to try to move away from > this library, especially if they changed to a Sage-uncompatible > license.
Here is the short story with Sage's c_graphs: I have done the work necessary for Sage to use them for just about everything in sight. I think there are a few functions which still depend on NetworkX, but this patch ( http://sage.math.washington.edu/home/rlmill/switch_on_cgraphs.patch ) switches the default for all graphs to use c_graphs instead of NetworkX -- it's old, and may need rebase. With this patch, almost nothing still uses NetworkX, but some things slow down a little. Although the underlying structure is faster, the translation in the middle is slowing things down. What someone ambitious should do is apply that patch, run their favorite function, and see whether it slows down or speeds up, and if it slows down, vet out why it does in the new backend code, to see what can be optimized. Everything is ready to switch, the only reason we haven't have been because of certain speed regressions (despite other functions getting faster ;) -- we'll just have to see what happens.) -- Robert L. Miller http://www.rlmiller.org/ -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
